In 2016, we analyzed violence-related fatalities of taxi and limousine industry drivers from 2003 through 2013 from the Bureau of Labor Statistics (BLS) Census of Fatal Occupational Injuries (CFOI). This analysis was conducted with restricted-access BLS datasets that are provided to the NIOSH Division of Safety Research under a memorandum of understanding. The BLS has assembled the CFOI annually to characterize all fatal work-related traumatic injuries in the United States since 1992. To identify and confirm fatalities, BLS uses multiple federal, state, and local sources such as death certificates, police reports, and workers' compensation reports. For a death to be recorded as “work-related” the decedent must have been employed at the time of the incident, working as a volunteer in the same capacity as a paid employee, or present at a site as a job requirement.¹⁵ CFOI includes all public and private sector workers regardless of age, but excludes fatalities occurring during a normal commute and deaths related to occupational diseases. Data fields are predominantly categorical with the exception of narrative text parsed into up to six text fields (depending on content) that may provide circumstances surrounding the fatality not elsewhere captured. Taxi and limousine service workers were identified by industry classification from the 2002 North American Industrial Classification System (NAICS) for 2003 through 2010,¹⁶ and the 2007 NAICS was used for 2011 through 2013.¹⁷ Although there were slight differences between the 2002 and the 2007 NAICS codes, the code for taxi and limousine service remained “4853” for both versions.

The Occupational Injury and Illness Classification System (OIICS) system developed by BLS classifies work-related injuries and illnesses into four separate hierarchical categorical structures: event/exposure, nature, primary and secondary source, and affected body part.^18,19 Each structure uses a four digit coding scheme that increases detail with each digit. The BLS OIICS defines nature of injury as the “principal physical characteristic(s) of the injury or illness” and the body part as that which is directly affected by the nature of injury or illness. Source and secondary source are the “objects, substances, equipment, and other factors that were responsible for the injury or illness incurred by the worker or that precipitated the event or exposure.” Event or exposure is “the manner in which the injury or illness was produced or inflicted by the source of injury or illness.”

We included the following variables about time, workplace and type of injury: time of incident, location of incident, establishment size, OIICS-coded variables for injury event, nature of injury, primary and secondary injury source, and body part injured. Establishment size refers to employer and is grouped into 1 to 10 employees,11 to 19, 20 to 49, 50 to 99, and 100 or more. Location of incident is characterized by the following examples: home, farm, mine/quarry, industrial structure, recreational location, street/highway, parking lot, and public buildings. We also included sex, age, race/ethnicity, nativity, and Bureau of Census (BOC) region. Sex was classified as either man or woman. Exact age at death was categorized into four major groups (less 35 yrs, 35 to 44 yrs, 45 to 54 yrs, and 55 yrs or older). Race/ethnicity was re-categorized into White (non-Hispanic), Black (non-Hispanic), Hispanic (any race), Asian, and all other races. Nativity was categorized into a dichotomous variable of either foreign born or US-born (including US territories, such as Puerto Rico, Guam, and the Virgin Islands). Region represented the four major US Bureau of Census regions: Northeast, Midwest, South, and West.²⁰ Major metropolitan areas with substantial taxi and limousine driver populations found in the Northeast are Boston, New York, and Philadelphia; the Midwest includes Chicago, Cincinnati, Columbus, and Detroit; the South includes Atlanta, Austin, Baltimore, Dallas, Houston, Miami, New Orleans, Orlando, Tampa, and Washington DC; and the West contains Denver, Honolulu, Las Vegas, Los Angeles, Portland, Reno, Sacramento, San Diego, San Francisco, and Seattle. The fatalities presented in this analysis are not limited to the metropolitan areas.

To identify the full range of violent fatalities, we included cases with one-digit event codes indicating “Assaults and Violent Acts” (Event code “6” for OIICS 1.01, 2003 to 2010) or “Violence and other injuries by persons or animals” (Event code “1” for OIICS 2.01, 2011 to 2013). This variable is classified as “intentional,” “unintentional,” or “intention unknown.” In order to identify the weapon used that resulted in the work-related fatality, we searched event, primary and secondary sources of injury for codes that indicated the type of weapon used, such as firearms, knives or cutting tools, and other types of weapons (multiple weapons, beatings, strangulation, or unknown weapon). We examined fatality narratives to classify the motive behind the violent fatality, including robbery, dispute, suicide, other and intentional violence by another person with unknown motive.

Rates were calculated using labor force denominator estimates derived from the BLS current population survey (CPS) for workers aged 16 years and older. The CPS is the principal source of US labor force statistics and is a monthly household survey that collects employment, unemployment, earnings, hours of work, and other indicators from a sampling frame of 60,000 households.²¹ Average annual rates were calculated as the total number of fatalities divided by estimated number of workers for all jobs for each of the characteristics of interest and are reported as the number of violence-related fatal work injuries per 100,000 workers. Unless otherwise stated categories with the lowest rates were used as the reference category.

We analyzed data from 2003 to 2013 since these years include the most recent data with consistent, comparable coding for race, ethnicity, and the taxi and limousine industry. For 2003 to 2013, we calculated the number, rate per 100,000 workers, and rate ratios for the study variables. Variable groupings were consistent with BLS confidentiality requirements and previous analyses. Poisson regression models were run in SAS using PROC GENMOD to calculate unadjusted (univariate) and adjusted (multivariate) fatality risk ratios and 95% confidence intervals (CIs). Rate ratios were adjusted for sex, age, race/ethnicity, nativity, and region. Data were analyzed in 2016 using SAS, version 9.3 (SAS Institute, Inc. Cary, NC).

No review was required by an Institutional Review Board as the analysis was conducted on existing data collected by other agencies and did not include any personal identifiers. The views expressed here do not necessarily reflect the views of the BLS.

We found 366 taxicab and limousine drivers died a violent death over an 11-year period at a rate of 17.8 per 100,000 workers (95% CI 16.9 to 18.7). Adjusted rate ratios revealed men died violent deaths 6.1 times the rate of women in this industry (Table 1) and violent death rates were higher for blacks (RR_adj 2.3 [95% CI 1.6 to 3.4]) and Hispanics (RR_adj 2.1 [95% CI 1.3 to 3.4]) than whites. Fatality rates were highest for the South (RR_adj 2.7 [95% CI 1.9–3.9]) and significantly higher for the Midwest (RR_adj 2.1 [95% CI 1.3–3.4]) and the West (RR_adj 1.8 [95% CI 1.2–2.8]) compared with the Northeast. Foreign-born taxi and limo drivers experienced significantly lower rates than US-born drivers (RR_adj 0.6 [95% CI 0.42–0.79]). Violent deaths were uniformly distributed across age groups (Table 1). Regardless of race/ethnicity the majority of fatalities occurred in establishments with less than 10 employees, on a local road or street, due to a robbery, by a firearm, or with the injury to the head, when reported (Table 2).

Where reported, a majority of violent deaths (42%) occurred in establishments with 10 or fewer employees (Table 2). A local road or street was the predominant location (71%). Robbery, when known, was the overarching motive (55%) and firearms (82%) the most frequently occurring weapon. The most frequently injured body part was the head, followed by the trunk.

Violent deaths among taxi and limo drivers predominantly occurred between 9 pm and before 7 am, with notable shifts in prevalence from nighttime occurrences to daytime occurrences (Fig. 1). The 24-hour distribution of violent deaths among all workers was more attenuated.

Each year, an average of 33 taxi and limousine drivers died a violent death with a range of 24 fatalities in 2009 to 47 fatalities in 2003. The fatality rates due to work-related violence among taxi and limousine drivers ranged from 8.7 per 100,000 workers in 2009 to as high as 21.7 per 100,000 workers in 2003. The rates were 14.5 (2009) to 21.7 (2003) times greater than all workers (Fig. 2). The rates were higher for US-born drivers compared with foreign-born drivers for every year examined, with the exception of 2003 to 2004 and 2013; these differences were not statistically significant. Black and Hispanic drivers had consistently higher violent death rates compared with white drivers (Fig. 3A). By contrast, violent death rates for all black workers were consistently higher compared with all Hispanic and all white workers (Fig. 3B).